It's a common question. Not everyone does, the camera has a good meter. Point&shooters won't bother looking at another meter, but many photographers can use a handheld incident meter. It still depends... The camera meter generally cannot meter flash (excepting one hot shoe TTL flash). To meter multiple flash in a studio situation, yes, the handheld meter is the necessary norm, to KNOW what your lights are doing. Photos outdoors without flash, it becomes greatly more optional. We all already use a light meter, a very good one is in the camera. However, it is a reflected meter (affected by subject colors), and the camera does Not meter manual flash. These are limitations. We can learn compensation to use the reflected meter in the camera, and then it does a lot. But a better handheld meter normally can meter incident or reflected light, and can meter flash or continuous light. An incident meter (metering principles) can certainly have strong advantages.
Reflected meters are aimed at the subject from the camera, and meters the light that the subject's colors reflect. The average of any reflected metered area is set to be about middle gray, 12.5% specifically for meters (gray cards are often 18%, which is 1/2 EV difference).
Said again, reflected meters can only try to place the scene's average tone near the middle.
A white dress or background reflects a lot of light, and reads high, so by moving this down to the middle, the meter underexposes the picture.
A black dress or background reflects little light, and reads low, so by moving this up to the middle, the meter overexposes the picture.
The expected reflected goal is that all results come out about middle gray brightness, not too dark, not too bright. This is all the meter can do, it just sees a blob of light, and does not know what the subject is, or how it should look. Fortunately, many typical scenes contain a random mix of dark and light colors that will probably average out about middle gray tone (not necessarily literally meaning gray, it might be all green or blue. Just meaning middle tone brightness, which is frequently referred to as middle gray). Then the middle tone result can often be about correct (because the purpose of the reflected meter is to put any result in the middle). When the scene is otherwise, we can compensate it, to correct it to come out as we want, brighter or darker, as needed.
A Spot Meter necessarily is a reflected meter, and then we need to know and consider the reflectivity of that selected spot if it is to have meaning. It does NOT AT ALL make the spot be "correct exposure", it just makes it come out about middle tone. It could be correct if you selected a middle gray tone spot. But if spot metering on a Caucasian face, you better open up about one more stop, to make it lighter.
Incident meters are the reverse, aimed at the camera from the subject, which reads the incident light from the light source directly. It meters the actual light incident on the subject, at the subjects location, instead of the light reflected from the subject. Incident metering does not see the subject, it instead meters the actual light, which makes it independent of the variations of colors reflected by the subject. Incident works for whatever is there in that light. So then any subject's tone, be its colors light, dark, or middle, is shown as it is. Kind of a big deal. Incident meters have the accuracy that newbies imagine their reflected meters ought to have (but can't). This is not possibly done at the camera, incident meters meter the light incident on the subject (actually from the subjects position, which is all important due to the Inverse Square Law). However, it can be less convenient to meter at the subjects position instead of just pointing the camera at it. Incident metering is NOT point&shoot.
IMO, the subject of handheld meters is primarily about incident metering. The camera meter can do as well or better than any handheld reflected meter, but incident meters have a practical accuracy only dreamed of by reflected meters. And incident meters can meter flash.
Incident meters are accurate, and are simply wonderful in the studio, but metering from the subjects location is generally more awkward to use out in the field.
Incident metering, and/or Manual flash, are probably about the only reasons to choose a handheld meter. Very big reasons however. See a closer look at incident metering differences. Your style of metering at the camera has to change though. Incident meters are used AT THE SUBJECT, to meter the actual incident light on the subject.
The incident meter may be awkward outside in the world, since it is metered at the subject location (in the subjects light) instead of from the camera location. However, Incident metering is wonderful in the studio, because it will meter the flash, and the subject location is handily available, and incident metering is simply more accurate (because it is independent of the reflected subject colors). Lighting can be set up before the subject arrives, the subject is unnecessary for metering, only the light at their location matters. Metering each individual flash is used to set that flash power level, which sets a desired lighting ratio. It is good for the entire session, and same setup is easily exactly duplicated next time too. (One exception: a hair light really has to be adjusted by eye, on the subject).
As a very quick spot check of meter accuracy, bright sun should meter near EV 15 at ISO 100. This EV 15 is f/16 at 1/125, or Sunny 16 would say f/16 at 1/100 second (both at ISO 100). Or equivalent exposures, f/11 at 1/250 or 1/200 second, or f/8 at 1/500 or 1/400 second, etc. Maybe not necessarily precisely EV 15, bright sun could often seen at times be a bit less bright, as this will vary slightly with air clarity and angle, haze, time of day or season, latitude, etc. The meter indicates the precise value.
Other than TTL flash, the camera meter does Not meter manual flash. We can easily adjust one hot shoe flash in manual flash mode, done trial and error by eye on the rear LCD preview image, or with the histogram. So people may tell you "nah, you don't need no stinkin' meter", but they only know about their one flash situation. But what about multiple studio flash? The idea is to control every flash unit, to know each one is doing exactly what you want. Not everyone uses multiple studio lights, but if you do, you will certainly appreciate the handheld meter (you already know this, really no other way to do it consistently). Otherwise, how will you duplicate your last sessions lighting setup again? Lighting is really no longer merely about exposure, but is about lighting ratio, the relative power of each flash. Lighting ratio is very important, and the handheld meter makes it be easy to set each light to an exact value, both easy to do it, and in particular, to know what you are doing, and to be able to repeat it next time. The camera commander becomes insufficient, it does control two individual lights, but our setup may use four lights. And the handheld meter is incident metering where the commander is reflected. These are big differences, and a big deal.
This all seems pretty obvious, but usually none of it is ever mentioned in discussions about if handheld meter is useful. Many people may not understand why a handheld meter might be used. The camera meter is excellent, it meters reflective metering very well, surely it is the best reflective answer. The only problem is that it is reflective, fooled by subjects reflected colors. This handheld meter purpose is normally about incident metering, and/or about metering flash. There may be other minor considerations, but if flash or incident metering advantages were your concerns, then the handheld meter is the solution. If you only want another reflected meter, I'd think about that longer. If you use reflected metering, the camera meter is hard to beat (but there is still flash). It has advantages (internal meter sees same angle of view as the lens), and it can do Spot too. But yes, other features are available, notably incident and flash metering.
This Sekonic L-308S meter, was the simpler and less expensive Sekonic model. The current version of it is Sekonic L-308X-U Flashmate Light Meter $219 (USD). No frills, but very nice and all the basics. It can meter incident or reflected light, and it can meter flash or continuous ambient. IMO for studio use, it will do all that needs to be done. I don't detect any difference in the current ad, other than a 3 year warranty now.
Incident and reflected metering results likely vary slightly, because incident is directly reading the actual light source incident on the subject, and reflected is reading the reflection from whatever 40 degree wide view and colors that it sees. Two different situations, and this is why we use incident (but a camera cannot, because it cannot meter at the subjects location).
I also have a Sekonic L-358 (no longer in production). The current replacement is the L-478D, which has a touch screen. Sekonic LiteMaster Pro L-478D-U Light Meter, $369 (USD).
The L-358 is very nice, it's been a classic, and I like its wheel better than the up/down buttons. If I used it outdoors, I'd like its aperture preferred option too. To meter reflected light, you have to swap out its dome with a flat one (which becomes a 54 degree field). But I don't need its frills, which are simply wasted on me. I've learned to use the camera meter outdoors, and only use a handheld meter for multiple studio flash, and the L-308S is not missing anything for that.
For metering reflected light, this L-358 model replaces the dome with a flat disk provided instead. The L-308 model has a disk for flat subjects, but for reflected, it removes the dome, to use the bare sensor instead. But frankly, if we have incident metering, we use incident. And incident is ideal for metering flash (at the subject).
Use of a light meter:
I use the handheld incident meter mostly for multiple flash studio, used a few minutes each session to setup the lights. It's great to measure each flash individually, to set the power level of each light right, to know what they are doing, and that they are doing what we want them to do. And then we can repeat the setup easily again next time. Incident meters are wonderful (better accuracy than reflected meters). However, I've also used reflected meters forever, so ambient exposure, or flash compensation, with the camera meter is no mystery, still convenient for everything else (other than multiple manual flash). However, beginners will need to learn that reflected meters are just a guide, affected by what you point them at, and which don't always indicate precise proper exposure. The L-308S is also an incident meter, so exposure accuracy is not much issue then, but incident meters do have to meter from the subjects position. Experience with both types can teach a few things about metering exposure.
The L-308S has all the basics, and works great. Minimal in that it doesn't have the frills, but first class on the basics we actually need. There is no question about its accuracy (Specifications list a slightly less extreme range, but same +/- 0.1 EV accuracy as top of the line). It just does not have any of the extras seen in more expensive models. It simply meters the lights, and does what it needs to do, and does it very well, but no frills. I call them added frills, of no use to me, but perhaps workaround is an experience/skill thing. Others may consider them more essential, but simplicity and ease of use are major features too.
Among these missing frills NOT in this L-308S model are these features:
To easily see this, tweak a test exposure of something bright to exactly line up almost at the right 255 edge of the (RGB) histogram. Then intentionally underexpose exactly one stop. That's 50%, and yes, linear data ought to now line up at digital 128. But the gamma histogram is not linear, and it is NOT at 128, because gamma computes half to be at 186 (73%). It will be around 3/4 scale, but it varies, because the camera is also doing other things too, white balance and contrast and Vivid, etc. Repeat this shot with Daylight and then Incandescent White balance, and compare the curves. The data edge probably won't be 186 in either, and ballpark is the best hope. Do you really want to calibrate on this? The histogram is NOT a light meter, and the digital histogram is not linear (it is gamma encoded).
If in dozens of different situations, the meter is always consistently off a little, and exactly predictable, that would be a reason to adust it. But certainly NOT based on the bad logic of ONE reading on a gray card! If you do want the meter to read 1/3 stop higher, you can just enter its ISO to be 1/3 stop lower than it is. But my bet is that the meter works great as it is, so be careful not to screw it up. My own personal metering variations are my problem (caused by me), and I'm guessing your problem too. There is some skill involved to learn to meter more consistently, especially for studio flash. The meter is better than I am.
Perhaps that seems a long list of omissions, and your needs may be different, and the L-358 does add these frills, but speaking of a studio situation, IMO it's just clutter, none of which helps me to meter the lights. After using the L-308S several years (for studio flash), I don't feel I'm missing anything. What the meter does do is to meter the light, very well. The L-308S does exactly what I want it to do, no more, no less. Simplicity and ease of use seems a major feature. I have both, and normally use the L-308S. However, the Sekonic L478D is about $100 more, about like the L358 except with a touch screen. It does include these other features. You can study the user manuals, which are available on the Sekonic product pages.
The L-308S is simple to use, no complex menus or features. It just measures the light. The meter has a mode for continuous light (the sun icon), and it has two modes for flash, called corded and cordless. Corded means using a PC sync cord as described in the example next below. Or Cordless uses no PC cord, but instead after pushing its button, it waits 90 seconds to see a flash fire, any way you can trigger it, and then it meters that flash.
It meters incident light with the dome aimed at the camera from the subjects position, and it meters reflected light with the dome aimed at the subject from the camera position. The Lumisphere (dome shown) is used for normal 3D subjects, and there is a flat "Lumidisc" used for metering flat 2D copy (paper, paintings, etc). The dome slides away from its lens only for purpose of swapping these two. The meter does not necessarily meter the same view that the lens will see. The L-308S meter's reflected field of view is 40 degrees, which field width matches a 50 mm FX lens, or a 32 mm DX lens. The L-308S uses one AA battery, which seems handy.
A light meter is used for real actual old fashioned manual flash, like studio flash units, or the mode called M on the speedlight. Specifically, no commander, no TTL. We cannot meter TTL flash, which is pointless anyway (the camera controls it), then all we meter is the weak TTL preflash. And the Commander firing all its flashed commands is very bad news too (some ifs and buts on the AWL page here). The light meter is for real manual flash.
I suggest for bigger cameras (DSLR, etc), to set Manual flash mode and Manual camera M mode, with Low ISO (like ISO 100, and set Auto ISO OFF) and with f/stop up near f/8 (but f/8 will be excessive diffraction for a small compact camera). Camera M mode will allow setting shutter speed to be Maximum Sync Speed (typically around 1/200 second). The ISO 100/ f/8, 1/200 second settings will keep any ambient light level out of the picture. You can take one test picture with flashes off (meaning disable any trigger so they do not flash) but with the necessary ambient light situation left on (even the flash modeling lights on if you will be using them). With flash level set to match the metered camera settings, the test picture result of these settings should be a nearly BLACK frame in the picture taken with flash disabled. That black picture means the ambient light that your eyes easily see is kept out by the settings, at insignificant level to affect your white balance or your lighting ratio, and with ample depth of field for a human portrait subject. That is all good, and the flash has been metered and setup to be perfect at these same settings.
This calculator relates flash power changes by mapping EV to percent change, or percent change to EV, both relative to the previous power level. So a value 50% does NOT mean setting TO 50% power level, here it means a change to 50% of the previous power. Same as EV means a change amount. The 50% of the previous power means like from 1/4 to 1/8 power level is half, and half power is -1 EV. Knowing that half of the previous power is 1 EV is useful when adjusting power level with a meter. And I normally meter multiple flash in tenth stops, for the lighting ratio reason shown next.
From the subject's location, I aim the meter directly at the flash being metered for ratio. I should replace the dome with the flat disk to meter this way (to exclude light from other angles), but it seems unimportant if only one light is on, and I don't bother. The dome definitely is aimed at the camera for any incident reading for camera exposure. The idea of the spherical dome is that it properly factors the cosine response of lights at other angles.
With only one flash turned on at a time, I use the Sekonic L-308S meter by connecting a 15 foot PC sync cord (cord comes with the studio light) between the meter and ONE flash (the only one flash turned on). From under the subjects chin, a button on the meter triggers the flash and meters it. I start with the Main light, and set its power level to read 1/3 or 2/3 stop less than the fstop I want the camera to use (the fill light will add it back, depending on ratio). Then I move the cord to the next of four lights (and turn only it on), and repeat to set its level. This is to set each light to the power I want it to be, so I know what they are doing. Like maybe main near f/8, fill 1.0 or 1.3 stops less than main, background often about same as main, and hair light depends on hair color, maybe 1 stop more than main for black hair, maybe 1 stop less than main for light hair. Then I know what I am doing, and I know what the lights are doing. And another overwhelming advantage, I can easily repeat this exact same setup quickly next time, no guessing. I can do all this (except hair) to be ready and tested before the subject arrives.
The Incident meter is aimed at each light for this, from subjects position (which could be me, sitting there). Background light is metered at background surface, aimed at its light. Hair metered at hair, aimed at its light. Hair light is a starting point, can be close, but hair color may require some trial and error (how it shines in a test picture).
Then I turn on only the two main and fill lights (no background, no hair light, etc) and meter this exposure for the camera aperture. Meter under subjects chin, aimed at camera here. You can be this test subject, metering under your own chin in the subject position.
Your subject may not have even arrived yet, but you're done and ready now (except for seeing the hair light). Just turn all lights on, move sync cord from meter back to camera, and go at it. For optical slaves, this cord connects to the light that is aimed at the others, which normally is the fill light, back near the camera, behind and above camera in my case. From the camera, fill light is aimed at the subject, but which is also towards the optical slaves on the other lights. This is so simple, and very reliable.
If you are using multiple manual flash, there are not many options... such an handheld meter is the way to meter flash, and to KNOW what you are doing (and to repeat it easily next time).
If you want to work in continuous light, the meter will do that too. As will the camera meter, but incident has strong advantages. So it may be optional, but the handheld metering subject is about incident or reflected. And incident means metering the actual light source directly, exposure independent of the subjects influence, but metered at the subjects location (with only minor exception, clear sky bright sun is mostly the same everywhere in sight).